I have a similiar gap (~25GB) and my understanding of this goes along the lines of:
Directories are typically represented as linked-lists of some description. They often grow, but never shrink. They are effectively files that change whenever you change an entry in that directly. They have a minimum size on disk (depending on your file system). My top-level directories in my cpool are all ~150kb, some 4096 top-level directories - thats 600MB right there. All of us have something like this somewhere. The directory "maps" (if you like) that contain the structure of each backup is stored under the "pc" directory in the root "data" directory. They have to be on the same partition as your "cpool"/"pool" or hardlinks don't work. Every backup of every PC consists of a number of directories. My backup summary page tells me I have some ~512 full backups and around 2000 odd incrementals, plus I have at least 50+ hosts in there, all with some logs, often 50MB+ of logs per host. Each one of the backup sets contains full directory maps (the file system is the database). It works fast, but the minimum size on disk for many of those directories is 4k (in my case). Additionally, each of those directories contains an attrib file, which has the ACLs and other bits. Anyway, if each of those structures (per host/backup) adds up to ~10MB of directories and attrib files, 2000 times 10mb is in the ball park of making up for my "missing" ~25GB. There are certain hosts (mailspools using maildir) that are going to aggravate this issue (lots of large directories with lots of small files!). I'm pretty sure that the performance penalty for using the tail-packing option on a file system such as ReiserFS would lead to misery and pain (ie I don't know about everyone else, but I have enough problems making my backups fit inside a 24 hour period without further impacting disk IO). I'm pretty sure I only need ~1000 directories (or less, if I take the attributes into account) at 10kbyte each (which seems typical looking at my directories) per backup (I have 2000+ individual backups) - 2000 x 1000 x 10kb =~ 20GB my 2c :) (actually, looks a bit more like 20c don't it). //Chris ----- Original Message ----- From: "Saturn2888" <[email protected]> To: [email protected] Sent: Sunday, 19 September, 2010 9:27:00 PM Subject: [BackupPC-users] Why is the BackupPC pool size 30GB less than the size on the Pool is 388.93GB comprising 2139798 files and 4369 directories (as of 2010-09-19 02:38), Pool file system was recently at 77% (2010-09-19 08:23) But df -h shows: /dev/mapper/vg-backuppc 576G 420G 128G 77% /var/lib/backuppc The volume group is only for BackupPC's pool, nothing else. There's a pretty large difference between 390GB and 420GB. +---------------------------------------------------------------------- |This was sent by [email protected] via Backup Central. |Forward SPAM to [email protected]. +---------------------------------------------------------------------- ------------------------------------------------------------------------------ Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing. http://p.sf.net/sfu/novell-sfdev2dev _______________________________________________ BackupPC-users mailing list [email protected] List: https://lists.sourceforge.net/lists/listinfo/backuppc-users Wiki: http://backuppc.wiki.sourceforge.net Project: http://backuppc.sourceforge.net/ ------------------------------------------------------------------------------ Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing. http://p.sf.net/sfu/novell-sfdev2dev _______________________________________________ BackupPC-users mailing list [email protected] List: https://lists.sourceforge.net/lists/listinfo/backuppc-users Wiki: http://backuppc.wiki.sourceforge.net Project: http://backuppc.sourceforge.net/
